Human induced variability during failure mode effects analysis

Abstract

Failure Modes Effects Analysis (FMEA) has been utilized in engineering since the 1940s with the main goal of identifying failure modes through systematic evaluation of a system, product, or process. This method has been widely adopted across many industries and provides valuable input to more detailed analyses such as failure analysis, risk analysis, root cause analysis, and design improvement. FMEA also forms part of the Design for Reliability (DfR) process when utilized early during product design. The approach is utilized extensively in methodologies such as Reliability Centered Maintenance (RCM) and provides valuable input to Condition Based Maintenance (CBM) programs — both important elements within product sustainment and design processes Although FMEA can be an effective tool within Reliability and Maintainability, the results of this approach are often limited by its susceptibility to human errors that can degrade the quality of an analysis. Data overload, wherein an analyst becomes overwhelmed by the sheer volume of failure modes to be evaluated, potential biases toward extreme severity ratings, the variance in team dynamics, individual past experience (operator versus designer), and years of experience have all been identified as possible sources of analysis error. Researchers have proposed FMEA quality improvement approaches ranging from simple solutions, such as best practices, to the application of fuzzy logic. While these methods can potentially improve FMEA quality, they offer no solid quantification of human error within the process. Specifically, potential subjectivity can be present during the analysis process when severity rankings are chosen, and superfluous information can bias the analysis. This paper will present initial results from a recent study that investigated sources of variation in FMEA analyses due to variances in the human decision making process. The study specifically investigated the impact of data availability (failure and mishap) and analyst experience on several FMEA variables, e.g., severity and mitigation selection. The paper will provide valuable insight into the improvement of the FMEA process, thus improving the reliability and maintainability in a technology-reliant world.